ja4-platform/tests/vm/Vagrantfile

# -*- mode: ruby -*-
# vi: set ft=ruby :
# =============================================================================
# Vagrantfile — VMs de test ja4ebpf multi-distro
#
# 3 VMs pour les tests unitaires eBPF sur kernel réel :
#   - centos8    : CentOS 8 (el8)
#   - rocky9     : Rocky Linux 9 (el9)
#   - rocky10    : Rocky Linux 10 (el10)
#
# Fournit un environnement kernel complet pour les tests eBPF :
#   - tracefs / debugfs montés
#   - perf_kprobe PMU disponible
#   - uprobes fonctionnels avec accept4 tracepoints
#
# Prérequis (host Ubuntu) :
#   sudo apt-get install -y libvirt-daemon-system libvirt-clients qemu-kvm libvirt-dev ruby-dev
#   vagrant plugin install vagrant-libvirt
#   sudo usermod -aG libvirt,kvm $USER  # puis se reconnecter
#
# Utilisation :
#   vagrant up                  # créer + provisionner toutes les VMs
#   vagrant up rocky9           # créer une seule VM
#   vagrant ssh rocky9          # connexion SSH
#   make test-vm-nginx          # test nginx sur Rocky 9 (défaut)
#   make test-vm-all            # tous les tests sur Rocky 9
#   ./tests/vm/run-all-vms.sh  # tests sur les 3 VMs
#   vagrant destroy -f          # détruire toutes les VMs
# =============================================================================

Vagrant.configure("2") do |config|

  # ── Désactiver synced_folder par défaut ─────────────────────────────────────
  config.vm.synced_folder ".", "/vagrant", disabled: true

  # ── Provider libvirt commun ─────────────────────────────────────────────────
  config.vm.provider :libvirt do |v|
    v.cpus       = 2
    v.memory     = 2048
    v.nested     = false
    v.cpu_mode   = "host-passthrough"
    v.driver     = "kvm"
    v.disk_bus   = "virtio"
    v.nic_model_type = "virtio"
  end

  # ── Synchronisation du projet via rsync ─────────────────────────────────────
  config.vm.synced_folder "../..", "/ja4-platform",
    type: "rsync",
    rsync__exclude: [".git/", "old/", "*.rpm", "dist/"]

  # ═══════════════════════════════════════════════════════════════════════════
  # VM 1 : CentOS 8 (el8)
  # ═══════════════════════════════════════════════════════════════════════════
  config.vm.define "centos8", autostart: false do |node|
    node.vm.box = "centos/8"
    node.vm.network "private_network",
      libvirt__network_name: "ja4-e2e",
      type: "dhcp"
    node.vm.provision "shell", path: "provision-el8.sh"
    node.vm.post_up_message = "VM centos8 prête ! Tests : make test-vm-centos8"
  end

  # ═══════════════════════════════════════════════════════════════════════════
  # VM 2 : Rocky Linux 9 (el9) — VM par défaut
  # ═══════════════════════════════════════════════════════════════════════════
  config.vm.define "rocky9", primary: true do |node|
    node.vm.box = "generic/rocky9"
    node.vm.network "private_network",
      libvirt__network_name: "ja4-e2e",
      type: "dhcp"
    node.vm.provision "shell", path: "provision.sh"
    node.vm.post_up_message = <<~MSG
      VM rocky9 prête !

      Depuis la racine du projet :
        make vm-ssh               # connexion interactive
        make test-vm-nginx        # test nginx complet (L3/L4 + TLS + L7)
        make test-vm-all          # tous les tests
        make vm-rebuild-ja4ebpf   # resynchroniser + recompiler après modif
    MSG
  end

  # ═══════════════════════════════════════════════════════════════════════════
  # VM 3 : Rocky Linux 10 (el10)
  # ═══════════════════════════════════════════════════════════════════════════
  config.vm.define "rocky10", autostart: false do |node|
    node.vm.box = "almalinux/10"
    node.vm.network "private_network",
      libvirt__network_name: "ja4-e2e",
      type: "dhcp"
    node.vm.provision "shell", path: "provision.sh"
    node.vm.post_up_message = "VM rocky10 prête ! Tests : make test-vm-rocky10"
  end

  # ═══════════════════════════════════════════════════════════════════════════
  # VM 5 : Traffic Generator (curl-impersonate + httpx)
  #
  # VM dédiée à la génération de trafic vers les endpoints.
  # Séparée des VMs endpoint pour des TLS fingerprints réalistes
  # et des IPs sources distinctes.
  # ═══════════════════════════════════════════════════════════════════════════
  config.vm.define "traffic", autostart: false do |node|
    node.vm.box = "generic/rocky9"
    node.vm.network "private_network",
      libvirt__network_name: "ja4-e2e",
      type: "dhcp"
    node.vm.provider :libvirt do |v|
      v.cpus   = 2
      v.memory = 1024
    end
    node.vm.provision "shell", path: "provision-traffic.sh"
    node.vm.post_up_message = "VM traffic prête ! Génération de trafic vers les endpoints."
  end

  # ═══════════════════════════════════════════════════════════════════════════
  # VM 4 : Analysis Server (ClickHouse + bot-detector + dashboard)
  #
  # VM centralisée pour le test E2E distribué. Les endpoints EL8/9/10 envoient
  # leurs logs ja4ebpf vers le ClickHouse de cette VM (192.168.42.10).
  # ═══════════════════════════════════════════════════════════════════════════
  config.vm.define "analysis", autostart: false do |node|
    node.vm.box = "generic/rocky9"
    node.vm.network "private_network", ip: "192.168.42.10",
      libvirt__network_name: "ja4-e2e",
      libvirt__netmask: "255.255.255.0"
    node.vm.provider :libvirt do |v|
      v.cpus   = 2
      v.memory = 8192    # 8 Go — torch + isotree build
    end
    node.vm.provision "shell", path: "provision-analysis.sh"
    node.vm.post_up_message = <<~MSG
      VM analysis prête !

      Depuis la racine du projet :
        make test-e2e       # test E2E complet (capture + ML + dashboard)
        make test-e2e-quick # test rapide avec trafic réduit
    MSG
  end

end